Fully Convolutional Architectures for Multiclass Segmentation in Chest Radiographs

• Published in: IEEE transactions on medical imaging Vol. 37; no. 8; pp. 1865 - 1876
• Main Authors: Novikov, Alexey A., Lenis, Dimitrios, Major, David, Hladuvka, Jiri, Wimmer, Maria, Buhler, Katja
• Format: Journal Article
• Language: English
• Published: United States IEEE 01.08.2018; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Architecture; Artificial neural networks; Biomedical imaging; Chest; chest radiographs; Clavicle - diagnostic imaging; clavicle segmentation; Databases, Factual; Feature extraction; fully convolutional network; Heart; Heart - diagnostic imaging; heart segmentation; Humans; Image classification; Image processing; Image segmentation; imbalanced data; JSRT dataset; Lung; Lung - diagnostic imaging; Lung segmentation; Lungs; Masks; Medical imaging; multi-class segmentation; Neural networks; Neural Networks (Computer); Object recognition; Organs; Radiographic Image Enhancement - methods; Radiographs; Radiography; Radiography, Thoracic - methods; Regularization; Shape; Task analysis
• ISSN: 0278-0062; 1558-254X; 1558-254X
• DOI: 10.1109/TMI.2018.2806086

The success of deep convolutional neural networks (NNs) on image classification and recognition tasks has led to new applications in very diversified contexts, including the field of medical imaging. In this paper, we investigate and propose NN architectures for automated multiclass segmentation of anatomical organs in chest radiographs (CXRs), namely for lungs, clavicles, and heart. We address several open challenges including model overfitting, reducing number of parameters, and handling of severely imbalanced data in CXR by fusing recent concepts in convolutional networks and adapting them to the segmentation problem task in CXR. We demonstrate that our architecture combining delayed subsampling, exponential linear units, highly restrictive regularization, and a large number of high-resolution low-level abstract features outperforms state-of-the-art methods on all considered organs, as well as the human observer on lungs and heart. The models use a multiclass configuration with three target classes and are trained and tested on the publicly available Japanese Society of Radiological Technology database, consisting of 247 X-ray images the ground-truth masks for which are available in the segmentation in CXR database. Our best performing model, trained with the loss function based on the Dice coefficient, reached mean Jaccard overlap scores of 95% for lungs, 86.8% for clavicles, and 88.2% for heart. This architecture outperformed the human observer results for lungs and heart.